Pith. sign in

REVIEW 1 major objections 4 minor 9 references

On the cohomology of line bundles over certain flag schemes

T0 review · 1 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read For a family of line bundles on the SL_{d+1} partial flag scheme, all nonzero cohomology is the kernel and cokernel of a single explicit multiplication map.

desk verdict The main cohomology computation is solid and the Proctor-determinant consequences are real, but the SL3 section contains an unproved free-rank claim that needs attention before the n>m description is complete. read the letter →

arxiv 1908.08438 v4 pith:X737DHXN submitted 2019-08-22 math.RT math.CO

Linyuan Liu This is my paper · ORCID
classification math.RTmath.CO MSC 05E1014L1520G05
keywords cohomologyoflinebundlesflagschemespartialvarietiesWeylmodulesmultinomialcoefficientscharacteristicpSL_{d+1}
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper computes the cohomology of the line bundles $L(m\omega_1-(n+d)\omega_d)$ on the $\mathbb{Z}$-scheme $G/Q$ of partial flags $D_1\subset H_d\subset V$ for $G=\operatorname{SL}_{d+1}$. It proves that only the two middle cohomology groups survive: $H^{d-1}$ is the kernel and $H^d$ is the cokernel of multiplication by the invariant form $f=X_0Y_0+\cdots+X_dY_d$ acting between two explicitly described free $\mathbb{Z}$-modules. It then identifies each weight space of $H^d$ with the cokernel of a matrix whose entries are multinomial coefficients, and on the diagonal $m=n$ it evaluates the determinant of that matrix by a product formula. From the determinant it derives torsion restrictions in characteristic $p$ and, in some ranges, identifies modular cohomology modules as simple modules. The upshot is a complete, integral, weight-by-weight description of these cohomology groups rather than a character or dimension count alone.

What carries the argument

The load-bearing object is the $\mathbb{Z}$-linear map $\varphi_{m,n}: S_{m-1}\otimes\Delta_{n+d+1}\to S_m\otimes\Delta_{n+d}$ given by multiplication by $f=X_0\otimes Y_0+\cdots+X_d\otimes Y_d$, where $S$ is the symmetric algebra on the standard representation and $\Delta$ is the explicitly defined module of inverse polynomials. Geometrically, $f$ is the defining equation of $G/Q$ as a hypersurface in the product of two projective spaces, so the cohomology of $L(\mu)$ is computed from the Koszul-style exact sequence of sheaves displayed in the paper. The argument reduces the sheaf cohomology to $\ker(\varphi)$ and $\operatorname{coker}(\varphi)$, and then a change of basis on the source turns the weight-space matrix of $\varphi$ into a block matrix whose only nontrivial block $M$ has multinomial coefficients $\binom{m-k}{b_1-b'_1,\ldots,b_d-b'_d}$. On the wall $m=n$, $M$ is square and its determinant is evaluated by the product formula in the paper's Proposition 3, which is what gives the torsion and simplicity statements.

What would settle it

Take a small case such as $d=3$, $m=n=1$ and compute the sheaf cohomology $H^i(G/Q,L(\mu))$ by a direct spectral sequence or computer algebra system: the predicted $H^1(Q/B,\mu)$ must vanish, and the rank of the cokernel of the explicit multinomial matrix must match the dimension of the corresponding weight space obtained from the classical character formula over $\mathbb{Q}$. Any mismatch in rank, or any nonzero $H^1(Q/B,\mu)$, would disprove the paper's description.

Watch

Extended reading notes

Core claim

Stated on the paper's own terms: for $\mu=m\omega_1-(n+d)\omega_d$, the only nonvanishing cohomology groups of $L(\mu)$ on $G/Q$ are $H^{d-1}(\mu)\cong\ker(\varphi_{m,n})$ and $H^d(\mu)\cong\operatorname{coker}(\varphi_{m,n})$, where $\varphi_{m,n}$ is multiplication by $f=X_0\otimes Y_0+\cdots+X_d\otimes Y_d$ from $S_{m-1}\otimes\Delta_{n+d+1}$ to $S_m\otimes\Delta_{n+d}$; here $S_r$ is the degree-$r$ piece of a polynomial ring and $\Delta_r$ is the degree $-r$ piece of a module of inverse polynomials. The map $\varphi$ is $G$-equivariant, so both cohomology groups carry natural $\operatorname{SL}_{d+1}$-actions. The proof realizes $G/Q$ as the hypersurface $V(f)$ inside $\mathbb{P}(V^*)\times\mathbb{P}(V)$ and shows that a long exact sequence collapses, leaving only $\varphi$. Weight by weight, $\operatorname{coker}(\varphi)$ is the cokernel of a matrix with multinomial coefficients indexed by two explicit sets $C$ and $D$; when $m=n$ the matrix is square and its determinant is given by a product formula. The determinant has direct consequences: for primes $p>n$ the group has no $p$-torsion, and in characteristic $p$ the special module $H^d_K(p,0,\ldots,0,-p-d)$ is the simple module of highest weight $(0,p-2,1,0,\ldots,0)$ with one-dimensional weight spaces.

Load-bearing premise

The entire calculation stands on the identification $H^i(\mu)\cong H^i(G/Q,\mu)$, which requires that the line bundle restricted to the fibres $Q/B$ have no cohomology beyond $H^0(Q/B,\mu)\cong\mu$; if $H^1(Q/B,\mu)$ did not vanish for this weight, the exact sequence (11) would gain extra terms and the kernel/cokernel description of the cohomology would fail.

Editorial extensions

If this is right

  • For every $m,n$, every weight of $H^d(m,0,\ldots,0,-n-d)$ lies below an explicitly computed dominant weight, and the weight spaces are cokernels of explicit multinomial-coefficient matrices with no hidden higher cohomology.
  • When $m=n$, the absolute determinant of the weight-space matrix is a product of factorials and rising factorials, so primes $p>n$ cannot divide the torsion; in particular $H^d(n,0,\ldots,0,-n-d)$ is $p$-torsion-free for $p>n$.
  • Over a field of characteristic $p$, $H^d_K(p,0,\ldots,0,-p-d)$ is the simple module $L_K((0,p-2,1,0,\ldots,0))$, whose weight spaces all have dimension one.
  • For $\operatorname{SL}_3$, the results give explicit binomial matrices for every weight of $H^2(m,-n-2)$, and for $n=ap^d+r$ with $0\le r<p$ the module $H^2(n,-n-2)$ is the quotient of a Weyl module by a single simple submodule.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The same hypersurface-in-$\mathbb{P}\times\mathbb{P}$ mechanism should apply to other pairs $(G,Q)$ where a Levi subgroup contributes no higher cohomology to the restricted character; the kernel/cokernel description is likely a special case of a broader pattern that cohomology is supported on the defining hypersurface of the flag scheme.
  • The determinant formula's $p$-adic valuation, which the paper computes only in special ranges, probably follows a base-$p$ digit rule for multiplicities; checking Lucas-type behavior for the cokernel of the multinomial matrix would give a coarse description of all torsion exponents.
  • Because the cokernel matrices are indexed by lattice points in hypersimplices, their Smith normal forms could be studied combinatorially; that would refine statements like 'the determinant has $p$-adic valuation 1' into a full description of the isomorphism type of the finite abelian group.
  • An immediate testable extension is to fix $d$ and $p$ and let $m,n$ vary, comparing the $p$-primary components of $H^d_K(m,0,\ldots,0,-n-d)$ with the explicit matrices; the paper's Corollary 2 and Corollary 5 suggest the torsion should be periodic in $n$ with period a power of $p$.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

1 major / 4 minor

Summary. The paper studies the integral cohomology of line bundles L(μ) on the partial flag scheme G/Q for G = SL_{d+1} over Z, where Q is the parabolic subgroup corresponding to α2,...,α_{d-1} and μ = mω1 - (n+d)ωd. The main reduction identifies the nonzero cohomology H^{d-1}(μ) and H^d(μ) with the kernel and cokernel of the multiplication map φ_{m,n}: S_{m-1} ⊗ Δ_{n+d+1} → S_m ⊗ Δ_{n+d} (equation (11)). The paper then gives weight-space descriptions of H^d(μ) as cokernels of explicit matrices with multinomial coefficients (Propositions 1 and 2), evaluates the determinant on the wall m = n using Proctor's formula, and derives p-torsion and simplicity consequences. In the final section, the SL3 case is made explicit (Corollaries 6 and 7), and a characteristic-p exact sequence is proved (Proposition 5).

Significance. If correct, the paper gives a complete integral computation of the cohomology of a family of line bundles on a partial flag variety, with explicit multinomial matrices, torsion information, and applications to simple modules with one-dimensional weight spaces. The main computation is carried out with explicit bases, recursive basis changes, and lemmas with proofs, and the reductions via Borel-Weil-Bott, the universal coefficient theorem, and Proctor's determinant formula are standard. The identification H^i(μ) = H^i(G/Q, μ) after equation (11) is sound: μ has zero pairing with the Levi simple roots α2,...,α_{d-1}, so its restriction to Q/B has H^0(Q/B, μ) ≅ μ and vanishing higher cohomology. However, one statement in the SL3 n > m case, Corollary 7(2), is asserted without proof, and that gap is load-bearing for the claimed complete weight-space description in that case.

major comments (1)
  1. [§6.1, Corollary 7(2)] The assertion that for k < n - m the ν_{t,k}-weight space of H^2(m, -n-2) is isomorphic to Z^{min(t,k)-max(0,t-m)+1} is stated without proof. Proposition 2 (with d = 2) reduces the weight space to the cokernel of the matrix (27), and Corollary 7(2) gives the displayed matrix (34) only when k ≥ n - m; no argument is supplied for the 'otherwise' case. The free-rank formula does not follow from the displayed matrix or from Proposition 4, whose hypothesis h1 ≥ k is not established in this range. Since this is part of the stated complete weight-space description for n > m in the SL3 case, the proof must be completed or the statement must be clearly labeled as an unproved conjecture.
minor comments (4)
  1. [§5] After Corollary 1, the paper infers p-torsion statements from the determinant formula (29). This is valid because the relevant matrix is square with nonzero determinant, so the cokernel is finite of order equal to the absolute value of the determinant; it would be helpful to state this explicitly.
  2. [Proposition 5, proof] The proof of Proposition 5 for r = p - 1 applies translation functors to a weight μ that lies on the wall of the facet F. The paper only notes that μ belongs to F if r ≠ p - 1; please add the precise statement from [Jan03] II.7.6 ensuring that the translation functor and the isomorphisms T^μ_λ V(λ) = V(μ) and T^μ_λ H^2(w2·λ) ≅ H^2(w2·μ) remain valid on the boundary.
  3. [General presentation] The manuscript contains numerous typographical and OCR-style artifacts, including 'd47' in the exact sequence (37), 'Cer tain' in the title, 'effeicients' in §6.1, and French words 'si' and 'pour' in the proof of Proposition 5. These should be cleaned up before publication.
  4. [References] The reference [Sup83] is cited for the statement that the weight set of L_K(λ0) consists of all dominant weights ≤ λ0; please provide a precise theorem number or a translation of the relevant statement, since the reference is in Russian and the cited assertion is strong.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the cohomology computation is a self-contained matrix calculation relying on standard theorems and Proctor's determinant formula; self-citations are confined to supporting SL3 applications.

full rationale

The derivation is not circular. The central exact sequence (10)-(11) is obtained by tensoring the standard hypersurface ideal sequence for G/Q = V(f) ⊂ P(V*) × P(V) and applying Künneth; no cohomology group is used as an input to define the matrix φ. The identification H^i(μ) = H^i(G/Q, μ) after (11) is valid because μ = mω1 − (n+d)ωd has zero pairing with the Levi simple roots α2,...,α_{d-1}, so it extends to a character of Q, forcing H^0(Q/B, μ) = μ and H^i(Q/B, μ) = 0 for i > 0; this is a standard Leray argument, not an assumption of the conclusion. Propositions 1 and 2 are derived by explicit monomial bases and the change-of-basis Lemmas 1 and 2, with the cokernel matrix determined by the multinomial coefficient of f; the wall determinant is taken from Proctor's independent formula and then evaluated by elementary p-adic valuation. The self-citations [Liu19] and [LP19] appear only in Section 6.2, where [Liu19] supplies a filtration for the SL3 application; they are not used to prove the main partial-flag matrix description, and [LP19] is mentioned only as an extension. Corollary 7(2) contains an unproved rank formula for k < n−m, but that is a missing proof or potential error, not a circular reduction. No fitted parameter is renamed as a prediction, and no input is defined in terms of the target cohomology.

Assumptions & free parameters 0 free parameters · 6 assumptions · 0 invented entities

No free parameters and no invented entities. The argument is a derivation from standard theorems; the external results are Borel-Weil-Bott, universal coefficients, Proctor's determinant formula, Doty's submodule structure, and Suprunenko/Seitz weight descriptions.

assumptions (6)
  • standard math Borel-Weil-Bott theorem over Q and the Weyl character formula
    Used to compute the rational cohomology H^d_Q(μ) and the free part of H^d(μ), in Section 2.
  • standard math Universal coefficient theorem for group cohomology (Jan03 I.4.18)
    Relates integral cohomology to cohomology over fields in equations (12)-(15).
  • standard math Proctor's determinant evaluation ([Pro90] Cor.1)
    Computes the determinant of the wall matrix in Proposition 3 and Corollary 1.
  • standard math Doty's description of the submodule structure of H^0(m,0) for SL3 ([Dot85])
    Used in the proof of Proposition 5, Section 6.2, to identify the kernel K.
  • standard math Suprunenko's and Seitz's descriptions of weights of simple modules ([Sup83], [Sei87])
    Used in Corollaries 3-5 to identify simple modules and one-dimensional weight spaces.
  • domain assumption The pushforward identification H^i(μ)=H^i(G/Q,μ) via H^0(Q/B,μ)=μ and vanishing higher cohomology
    Stated in Section 2 after equation (11); it is load-bearing and depends on μ being trivial on the Levi simple roots.

how reviews work

0 comments
Cite this review

Pith. "Pith review of On the cohomology of line bundles over certain flag schemes." pith.science (2026). https://pith.science/paper/X737DHXN

@misc{pith2026190808438,
  author       = {Pith},
  title        = {Pith review of: On the cohomology of line bundles over certain flag schemes},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/X737DHXN}},
  note         = {Machine review of arXiv:1908.08438}
}
abstract

Let $G$ be the group scheme $\operatorname{SL}_{d+1}$ over $\mathbb{Z}$ and let $Q$ be the parabolic subgroup scheme corresponding to the simple roots $\alpha_{2},\cdots,\alpha_{d-1}$. Then $G/Q$ is the $\mathbb{Z} $-scheme of partial flags $\{D_{1}\subset H_{d}\subset V\}$. We will calculate the cohomology modules of line bundles over this flag scheme. We will prove that the only non-trivial ones are isomorphic to the kernel or the cokernel of certain matrices with multinomial coefficients.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

9 extracted references · 7 canonical work pages

  1. [1]

    Doty, The submodule structure of certain Weyl modules for groups of type An

    Stephen R. Doty, The submodule structure of certain Weyl modules for groups of type An. J. Algebra 95 (1985), no. 2, 373–383

    work page 1985

  2. [2]

    Second edition

    Jens Carsten Jantzen, Representations of algebraic groups. Second edition. Mathematical Surveys and Monographs, 107. American Mathematical Society, Providence, RI, 2003. xiv+576 pp. ISBN: 0-8218-3527-0

    work page 2003

  3. [3]

    Kempf, Algebraic varieties

    George R. Kempf, Algebraic varieties. London Mathematical Society Lecture Note Series, 172. Cambridge University Press, Cambridge, 1993. x+163 pp. ISBN: 0-521-42613-8

    work page 1993

  4. [4]

    The Andrews Fest\-schrift (Maratea, 1998)

    Christian Krattenthaler, Advanced determinant calculus. The Andrews Fest\-schrift (Maratea, 1998). Sém. Lothar. Combin. 42 (1999), Art. B42q, 67 pp

    work page 1999

  5. [5]

    arXiv:1903.08758 [math.RT]

    Linyuan Liu, Cohomologie des fibrés en droites sur SL3 /B en caractéristique positive : deux filtrations et conséquences. arXiv:1903.08758 [math.RT]

    arXiv 1903

  6. [6]

    Linyuan Liu and Patrick Polo, On the cohomology of line bundles over certain flag schemes II, submitted [to this journal]

    work page

  7. [7]

    Proctor, Product evaluations of Lefschetz determinants for Grassmannians and of determinants of multinomial coefficients

    Robert A. Proctor, Product evaluations of Lefschetz determinants for Grassmannians and of determinants of multinomial coefficients. J. Combin. Theory Ser. A 54 (1990), no. 2, 235–247

    work page 1990

  8. [8]

    Gary Seitz, The maximal subgroups of classical algebraic groups, Mem. Amer. Math. Soc. 67 (1987), no. 365

    1987

Show all 9 references
  1. [9]

    Irina D. Suprunenko, Preservation of systems of weights of irreducible representations of an algebraic group and a Lie algebra of type A_ with bounded [restricted] higher weights in reduction modulo p (Russian), Vestsi Akad. Navuk BSSR Ser. Fiz-Mat Navuk, 1983, no. 2, 18-22

    1983

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.